How to Treat Dog Separation Anxiety: A Step-by-Step Guide

Dog separation anxiety is treated in two parts. The first is desensitization and counterconditioning: you practice absences short enough that your dog stays calm, then lengthen them slowly. The second is never leaving the dog alone for longer than it can currently handle while that training takes hold. Moderate and severe cases often also need prescription medication from a veterinarian. The medication supports the training and does not replace it.

Most guides explain that protocol well. What they leave out is how to judge, week by week, whether the protocol is working, and when to bring in more help. That decision matters because anti-anxiety medications take weeks to reach full effect. If you wait until training has clearly stalled before you call your vet, you add that delay on top of the weeks you have already lost. This guide is built around one tool: an escalation table you check against video of your dog’s absences, which tells you whether to keep going, add a trainer, ask your vet about medication, or see a veterinary behaviorist.

The short version of the treatment itself

Every reputable source covers the basics, so here they are briefly:

• This is distress, not spite. Punishing a dog for what happens while you’re gone makes the anxiety worse. The AVSAB position statements cover why punishment-based methods are a poor fit for fear and anxiety problems.
• Rule out look-alikes first. Urinary problems, incomplete house-training, marking, boredom, and noise phobia can all leave a mess or a chewed door frame. A vet exam comes before any behavior plan.
• The core method is graded absences. You start with absences of a few seconds and lengthen them only while your dog stays relaxed. You pair departures with something good when that helps. The ASPCA’s behavior team describes this as the center of anxiety treatment, with medication added when the dog is too anxious to learn.
• Keep departures and returns boring. Exercise and enrichment help a dog settle, but they don’t fix separation anxiety on their own.

The rest of this article covers what to do with those basics in practice.

Step 1: Record a baseline before you change anything

You can’t tell whether treatment is working unless you measure it. Before your first training session, set up a phone or cheap pet camera and record two or three ordinary departures. Write down two numbers for each one:

1. Time to first distress sign. Count from the door closing to the first whine, pacing at the door, heavy panting, scratching, or howling.
2. What the dog does with food. Does it eat a treat or work a food toy after you leave, or ignore it?

These recordings are your baseline. Every checkpoint below compares new recordings against them. How the dog feels about your leaving is a guess, but the time on the video is something you can compare from week to week.

Step 2: Place your dog in a severity tier

Use your baseline video, not your impression from coming home. Many owners underestimate severity because they only see the aftermath.

Tier	What the recordings show	Start with
Mild	Some whining or pacing that fades on its own; the dog eats when you leave; no damage, no accidents	Graded absences alone; food toy allowed (see the rule below)
Moderate	Distress starts soon after you leave and keeps going; damage near exits; indoor accidents in a house-trained dog; food often ignored	Graded absences, plus no alone-time beyond the dog’s limit, plus a vet appointment booked in the first two weeks
Severe	Escape attempts at doors or windows; constant vocalizing; drooling, panting, or trembling; no settling at all	Vet visit before you start training; medication discussed from the beginning; a qualified trainer or behaviorist guiding the plan
Crisis	Any self-injury: broken teeth, bleeding paws or gums, torn nails; frantic escape through glass or crate bars	Same-day or next-available vet contact; the dog is not left alone at all until a plan is in place

If your dog falls between two tiers, treat it as the more severe one. Scaling back support costs you little. Under-treating costs you weeks.

Step 3: Run a plan short enough to follow

Long lists of instructions get abandoned. A review of separation anxiety treatment in Veterinary Medicine: Research and Reports summarizes findings that owners follow only part of complex behavior plans, and that more complicated instructions are linked to worse results. So each tier below has no more than five actions.

Mild: five actions

1. Run short absence sessions on most days, lengthening the absence only after a fully relaxed repetition.
2. Use a food toy at departures, following the rule below.
3. Keep leaving and coming back low-key: no long goodbyes and no excited greetings.
4. Give the dog a walk or sniffing session before practice.
5. Re-record a real departure once a week and compare it with the baseline.

Moderate: five actions

1. Do everything from the mild tier, except drop the food toy if it fails the rule below.
2. Arrange cover so the dog is never alone past its current limit (see the work section below).
3. Book a vet visit in the first two weeks to rule out medical causes and discuss medication. Don’t wait until training stalls.
4. Use a gated room or other safe area instead of a crate unless your recordings show the dog is calm in the crate.
5. Send your videos to a qualified trainer if the week-3–4 checkpoint shows no gains.

Severe or crisis: five actions

1. Have a vet exam and a medication discussion before training starts.
2. Find a board-certified veterinary behaviorist through the American College of Veterinary Behaviorists, or work with a certified separation anxiety trainer who coordinates with your vet.
3. Arrange zero alone time until the plan is in place.
4. Remove whatever the dog has injured itself on: crate bars, window access, gaps under doors.
5. Change only what the professional tells you to change.

The food-toy rule

Advice about stuffed food toys conflicts. Some guides recommend one at every departure. Others warn that in serious cases the toy itself becomes a “you’re about to leave” signal, and that the dog panics as soon as the food runs out.

Here is the rule to follow: use a food toy only for mild cases, and stop if your recordings show distress begins when the toy is finished. The video settles the question quickly. If the dog calmly works the toy and then naps, the toy is helping. If the dog works the toy, licks it clean, then heads straight to the door and starts whining, the toy is only delaying the panic. In that case, take it out of departures.

Step 4: Check progress against the escalation table

At each checkpoint, compare a recorded real-world departure and your training-session notes against the baseline. The timing builds in the medication delay: for moderate and severe dogs, the vet conversation happens early, so that if medication is needed it is already working by the time training would otherwise have stalled.

Checkpoint	What the recordings should show	Signs you’re stuck	Action
Baseline (before week 1)	Time to first distress sign and food behavior recorded for two or three departures	Any self-injury or escape attempts	Contact your vet now and skip the do-it-yourself stage
End of week 2	Calm time in training sessions is clearly longer than baseline; the dog eats when you step out	Even a few seconds out the door causes distress; the dog won’t touch food when you leave	Mild, making gains: keep going. Mild but stuck, or any moderate dog: book the vet appointment now and raise medication
End of week 4	Real departures show the first distress sign coming later than at baseline; setbacks recover within a session or two	Calm time flat for two straight weeks; good in sessions, but real departures still look like baseline	Add a qualified trainer to review your videos. Moderate dog not yet seen by a vet: go now
Weeks 5–8	Steady gains. A dog on medication should look noticeably easier to work with as the drug reaches full effect	On medication with good technique and still flat; anxiety spreading to new triggers such as car rides, the crate, or you moving between rooms	See a veterinary behaviorist (DACVB directory) for re-evaluation and a medication review
Week 9 onward	Absences long enough to cover short errands, then gradually stretched toward your real schedule	Two-week plateaus that keep recurring; regression after each schedule change	Behaviorist re-evaluation; review whether cover arrangements are actually preventing panic

The table only works if you also follow two rules:

• Two weeks flat counts as stuck. One bad day doesn’t, and neither does a dip after a thunderstorm. Two weeks with no gain does.
• Progress in sessions alone doesn’t count. If training-session numbers rise while real departures still look like the baseline, the dog has learned your practice routine, not how to be alone.

Worked example: a moderate case, start to week 4

Juno is a four-year-old, 45-pound shepherd mix. Her owner works a hybrid schedule: office days Tuesday through Thursday, from 8:15 a.m. to 5:45 p.m. door to door, which is 9.5 hours away.

Baseline. Three recorded departures show Juno’s first whine at 1 minute 50 seconds, 2 minutes 20 seconds, and 2 minutes 5 seconds, so about 2 minutes on average. She scratches at the door by about 6 minutes and ignores a stuffed food toy every time. Ignoring food and damaging the exit both put her in the moderate tier, and the food toy fails the rule.

Cover math. Juno’s current limit is under 2 minutes, and the office days need 9.5 hours. That is a 9.5-hour gap, three days a week, or 28.5 hours of cover weekly. Her owner books daycare on Tuesday and Thursday, and a neighbor who works from home takes Juno on Wednesday. On home days, errands are batched into one trip, and Juno rides along in the car, which she handles calmly.

Week 1. The owner runs a few short sessions a day. The first session starts with 10-second absences, and the owner moves up only after a relaxed repetition. By day 7, Juno’s best calm absence in a session is 1 minute 15 seconds. That is shorter than her 2-minute baseline, which is expected, because training absences stop before distress starts.

Week 2 checkpoint. Best calm session time: 2 minutes 30 seconds. On a recorded real departure (a 4-minute trip to the mailbox), her first whine came at 2 minutes 40 seconds, later than baseline but only slightly. Because Juno is moderate, the owner books the vet now rather than waiting. The vet rules out medical causes, discusses fluoxetine, and the owner decides to start it.

Week 4 checkpoint. Best calm session time: 6 minutes. On a recorded real departure, the first distress sign came at 5 minutes 10 seconds. This is a clear trend away from baseline, so the decision is keep going: no trainer needed yet. The medication started in week 2 is still building toward full effect. Had the owner waited until a week-6 stall to raise medication, Juno would only now be starting a drug that needs weeks to work.

What would have changed the plan: if Juno’s real departures at week 4 still showed a first whine around the 2-minute mark, the next step would have been sending the videos to a trainer and checking in with the vet about the dose.

Can separation anxiety in dogs be cured, or only managed?

Many dogs improve to the point where owners stop thinking of them as anxious. They can be left for a normal workday and settle quickly. Others improve a great deal but stay sensitive: a move, a new baby, or a long stretch of remote work can bring back some distress, which usually responds to a short return to the protocol. The honest answer is that outcomes vary from dog to dog. Plan for long-term management, and count it as a win if the dog no longer needs it.

How long does separation anxiety training take before I see results?

You should see direction within the first two weeks: training-session absences getting longer and real departures showing distress starting later. Reaching a full workday alone usually takes months, not weeks, and longer in severe cases. If you see no direction at all by the end of week 2, use the table. Don’t simply keep going and hope.

How long do Prozac (fluoxetine) or Clomicalm (clomipramine) take to work?

Both take weeks, not days, to reach full effect. Your vet will tell you how long to wait before judging the drug. They are different kinds of drug: fluoxetine is a selective serotonin reuptake inhibitor (SSRI), and clomipramine is a tricyclic antidepressant. Both are prescription-only and both have side effects. Reduced appetite and lethargy are among those commonly reported, and your vet will want to hear about them. Your vet may also prescribe a fast-acting medication such as trazodone or gabapentin for specific departures while the daily medication builds up.

The research supports medication alongside training. In a controlled trial summarized in the review linked above, dogs given fluoxetine plus behavior work improved more than dogs given behavior work plus a placebo. The placebo group also improved, which shows how much the training itself does.

This delay is why the escalation table moves the vet conversation earlier. For moderate and severe dogs, decide about medication in the first two weeks, not after months of slow progress.

Pheromone diffusers such as Adaptil are a common question. They’re low-risk, but the evidence is limited and mixed. Use one as an add-on if you like, never in place of training or a medication your vet recommends.

Will getting a second dog help my dog’s separation anxiety?

Usually not. Most dogs with separation anxiety are distressed about being apart from a particular person, not about being without company in general, so they panic with another dog in the house too. You also risk adding a second dog that picks up the stress. A quick test before committing: borrow a calm friend’s dog for a recorded practice absence. If your dog’s first distress sign comes at the same time as usual, a second dog won’t fix the problem.

Should I crate a dog with separation anxiety?

Only if your recordings show the dog is relaxed in the crate while you’re gone. For many anxious dogs, a crate adds a second panic, about being confined, and dogs trying to escape crates can break teeth and tear nails. A gated room with the dog’s bed, water, and nothing it can injure itself on is the safer default. If you already crate and the video shows the dog pressed against the door, biting the bars, or digging at the floor, stop using the crate for alone time.

What do I do with my dog when I have to go to work during training?

Treat it as a scheduling problem and solve it before training starts. Every absence longer than the dog can handle is practice at panicking, and it undoes the work you did in sessions. These are the options, roughly from most to least reliable:

Option	Works best for	Watch out for
Daycare	Social dogs that are comfortable with other dogs	Some anxious dogs find busy daycares stressful; check how your dog behaves at pickup
In-home sitter or dog walker who stays	Dogs that do best at home	It only works if they stay until you’re back; a 30-minute walk in the middle of the day leaves hours of alone time
Neighbor, friend, or family member	Dogs attached to you specifically but not to being home	Agree on a backup plan for when they cancel
Taking the dog along	Dogs that travel calmly; workplaces that allow dogs	Never leave the dog in a parked car as a substitute
Time off or remote-work weeks	Severe and crisis cases while medication takes effect	Plan how you’ll gradually return to your normal schedule

Cover costs are a real part of owning a dog with this problem. If you’re budgeting for daycare or sitters, it’s worth checking whether your pet insurance covers behavior consultations. Many policies don’t.

Common mistakes

• Judging progress by the state of the house when you get home. A clean house can hide an hour of whining. Use video.
• Moving up after one good repetition. Move up only after relaxed repetitions, and drop back after any sign of distress.
• Calling a session plateau “slow progress.” Two weeks flat means you’re stuck and should change something. Use the table.
• Waiting until training fails before raising medication. Medication has a built-in delay. For moderate and severe dogs, ask early.
• Leaving the dog alone “just this once” past its limit. One long panic can undo a week of sessions.
• Making the departure routine more elaborate. Long goodbyes, special phrases, and treat rituals can all become signals that you’re about to leave.

When to see a vet, and when to get professional help

Contact your vet the same day or the next available day if:

• Your dog has injured itself: broken teeth, bleeding paws or gums, torn nails, or cuts from escape attempts
• Accidents or destruction come with lethargy, vomiting, diarrhea, appetite changes, or increased thirst. These may be medical problems, and only a vet can sort that out
• Your dog won’t eat or drink at all when alone, even during very short absences

Book a vet appointment (not an emergency) if:

• Your dog is in the moderate tier or above. Book within the first two weeks of training
• The week-4 checkpoint shows no change from baseline
• You see side effects from a medication your dog is already taking

See a board-certified veterinary behaviorist (find one through DACVB) if:

• Your dog is in the severe or crisis tier
• Medication plus well-run training still shows no gains in weeks 5–8
• The anxiety is spreading to other situations, which can point to a broader anxiety problem that needs a different plan

Separation anxiety is a behavioral condition, and your vet is the person who confirms it and rules out everything else. This article helps you organize the work and decide when to escalate. It is not a substitute for that exam.

What success looks like on video

Eventually your recordings will look uneventful. The door closes, and the dog sniffs around, lies down, shifts position a few times, and sleeps until you come back. Getting there takes video, short honest sessions, cover arrangements for real-life absences, and checkpoints that tell you to escalate before the weeks slip by. If the checkpoints say you’re stuck, bringing in a trainer, a vet, or a behaviorist is the planned next step, not a sign that you failed.